I categorised all 360 VARC questions from CAT 2021–2025. One question type is 33% of the section.
RC Inference alone accounts for 33% of all VARC questions. The shift happened in 2022 and hasn't reversed. Here's the full breakdown — and why most VARC prep plans get the mix wrong.
The data
All 360 VARC questions from CAT 2021–2025, tagged by type and question subtype. Split: 67% Reading Comprehension, 33% Verbal Ability (Para Summary, Para Jumble, Sentence Insertion, Odd Sentence Out).
Topic breakdown
| Topic | Questions | % of VARC |
|---|---|---|
| Reading Comprehension | 240 | 67% |
| Para Summary | 39 | 11% |
| Para Jumble | 35 | 10% |
| Sentence Insertion | 22 | 6% |
| Odd Sentence Out | 18 | 5% |
| Para Completion | 6 | 2% |
Questions by topic (360 total)
Reading Comprehension — 240 questions (67%)
4 passages per slot, 4 questions each = 16 RC questions per slot. Here's the breakdown by question type across all 240 RC questions:
| Question Type | Questions | % of RC |
|---|---|---|
| Inference | 117 | 49% |
| Detail / Factual | 44 | 18% |
| Main Idea / Primary Purpose | 28 | 12% |
| Tone & Attitude | 12 | 5% |
| Vocabulary in Context | 11 | 5% |
| Critical Reasoning | 8 | 3% |
| Application / Extension | 7 | 3% |
| Structure / Purpose | 5 | 2% |
| EXCEPT-type | 5 | 2% |
The headline number: Inference is 49% of RC and 33% of the entire VARC section.
The shift that happened in 2022
Year-by-year Inference as a share of RC questions:
| Year | RC Questions | Inference | Inference % |
|---|---|---|---|
| 2021 | 48 | 12 | 25% |
| 2022 | 48 | 28 | 58% |
| 2023 | 48 | 21 | 44% |
| 2024 | 48 | 30 | 62% |
| 2025 | 48 | 26 | 54% |
RC Inference as % of RC questions — year by year
2021 was an outlier — close-reading Detail questions dominated. From 2022 onwards, Inference has been the majority question type every single year. If your mocks are based on 2019–2021 papers, you're practising the wrong mix.
Three things that surprised me
1. Inference is not just "important" — it's the single most tested skill in the entire CAT paper.
117 questions out of 1,002 total = 11.7% of the complete exam. Treating it as one RC type among many is a calibration mistake.
2. Para Jumble gets too much attention for how rarely it appears.
35 questions in five years = 2–3 per slot. It's declining year on year. Para Summary (39 questions) appears at the same frequency and is considerably more learnable — there's a clearer method, and the skill transfers to RC Main Idea questions.
3. Para Completion has effectively disappeared.
6 questions in five years. Some slots had zero. Not worth dedicated prep time.
Suggested time allocation (100 hours for VARC)
| Focus area | Hours | Why |
|---|---|---|
| RC — Inference specifically | 40 | 117 questions; single highest-frequency skill |
| RC — all other types | 20 | Detail, Main Idea, Tone |
| Para Summary | 15 | 39 questions; learnable pattern |
| Sentence Insertion | 10 | 22 questions; consistent presence |
| Odd Sentence Out | 8 | 18 questions; growing trend |
| Para Jumble | 7 | 35 questions; declining but still appears |
| Para Completion | 0 | 6 questions in 5 years |
The ratio most plans give VA vs RC is wrong. RC deserves ~75% of VARC hours, not 50%.
Put this into practice
Solve CAT VARC PYQs — filtered by topic, with full solutions.